Codex vs Zcode
قارن النماذج
Codex is more worth it — 58% vs 34%
How we score worth-it? →Reliable executor — lower ceiling, generous limits
- In my experience gpt5-codex (medium) and codex-cli is notably better than Sonnet 4.5 and claude-code. It is slower, but the results are much more often correct and it doesn't rush into half-baked solutions/dumb approaches as eagerly. I'd much rather wait 5 minutes than have to clean up manually.— hn
- I was able to use it for few use cases where it got it right the first time. The ability to use a screenshot to create an application is nice for rapid prototyping.— hn ↗
- Codex: a bit slow, but its instruction-following is quite good.— v2ex ↗
- By the time I moved to Codex, I felt the AI could handle UI design (frontend), implement features, and even test the project in the browser by itself.— v2ex ↗
- 我跟up主的感觉有点不一样,在我看还是Codex做事更稳一些。同样的提示词给到Codex和antigravity,出来的结果差异不小。Codex做事不急躁,稳扎稳打;而antigravity跟Gemini简直一个模子刻出来的,都是“谄媚+激进”。当然,这不是说antigravity不好,而是适合的任务与Codex不同。— bilibili ↗
- I've been quite unimpressed by Codex for now... the quality of the code is worse than Claude for me.— hn ↗
- I'm done. Codex was dumb as a rock today - on 6.1 it keeps stopping to report back, can't get through a single long task, and reports in again after every couple of file edits. Absolutely braindead.— v2ex ↗
- hi, just got 0.157.0, highlighting text in the TUI now has a pretty rough (~.5 seconds) latency. I copy the outputted text a lot, so this is a major pain point. Have to downgrade to 0.156.0 till this is fixed 😅— github ↗
“Great for kicking off background tasks and walking away”
“Predictable cost and generous limits — the budget pick for heavy use”
“But it feels lazy on complex reasoning. It'll take shortcuts you didn't ask for”
Z.ai's coding agent on GLM-5.2
- It found dead-end links, a long section without a table of contents, and a timezone inconsistency. That was useful work from the first task.— blog ↗
- It definitely can be a daily driver if you don't want to use Anthropic or OpenAI. It's going to be even better with native vision now available.— hn ↗
- Pretty impressed, it did good work. Good reasoning skills and tool use. z.AI offers a coding plan which is absolutely worth it.— hn ↗
- Right in the middle of a conversation the model suddenly switched from GLM 5-3 to GLM 5-3 HighSpeed, and the transition animation was pretty slick.— v2ex ↗
- zcode太良心了今天晚上又可以领取5亿token了!— bilibili ↗
- Whenever you are logged in, ZCode silently packages your entire workspace — complete .git history, LFS asset cache, reflogs, and global app configs — encrypts it, and uploads it to Aliyun OSS.— blog
- The pricing and promotion terms contradict each other, so the actual cost can't be accurately calculated right now.— blog ↗
- It looks like Z Code is just a Codex fork... when it comes to trusting my entire machine with it, I just run Codex.— hn ↗
- When writing a somewhat larger project, token consumption is way too fast — it's especially obvious with Zcode.— v2ex ↗
- I went back to Trae. I don't see any obvious advantage in Zcode.— v2ex ↗
“Z.ai's coding agent, built on Z.ai's own 5.2 base model”
“Rivals the top coding IDEs on the benchmarks”
“Strong in Chinese and English, which is rare”
Which should you pick?
The gap is clear: Codex sits at 58% while Zcode is at 34%, a 24-point difference. If you just want the safer pick, start with Codex.
On entry price, Zcode is cheaper ($18 vs $20, Codex costs about 11% more).
Both are in the وكيل category with similar coverage; the real difference is stability and cost — run one of your own tasks through each before deciding.
What users actually said about Codex: "In my experience gpt5-codex (medium) and codex-cli is notably better than Sonnet 4.5 and c…"
FAQ
Is Codex or Zcode more worth it?
Codex is more worth it, with a 58% worth-it score vs 34% for Zcode.
How much does Codex cost vs Zcode?
Codex: Free · Plus $20/mo · API $1.75/$14 · Zcode: From $18/mo.
What do users say about Codex vs Zcode?
Codex: Great for kicking off background tasks and walking away. Zcode: Z.ai's coding agent, built on Z.ai's own 5.2 base model.